Part VII — Chapter 20

Metabolic Engineering

Rational redesign of plant metabolic pathways — from Golden Rice and herbicide-resistant crops to CRISPR-edited secondary metabolite biosynthesis — guided by metabolic flux analysis and genome-scale stoichiometric models.

Metabolic Engineering Workflow

Major Case Studies

Golden Rice — Beta-Carotene in Endosperm

Wild-type rice endosperm accumulates GGPP but lacks the downstream enzymes for carotenoid biosynthesis. Golden Rice 2 (2005) expresses two transgenes to restore the pathway: phytoene synthase (PSY) from maize (OsGGPPS provides GGPP substrate) and carotene desaturase (CrtI) from Erwinia bacteria (performs all four desaturations to lycopene in a single enzyme). Endogenous plant lycopene cyclase then converts lycopene to beta-carotene. Result: up to 37 μg/g dry weight beta-carotene, a 23-fold increase over GR1.

\( \text{GGPP} \xrightarrow{\text{PSY (maize)}} \text{phytoene} \xrightarrow{\text{CrtI (}E.\text{uredovora)}} \text{lycopene} \xrightarrow{\text{LCY-b (endogenous)}} \beta\text{-carotene} \)

Humanitarian approval in Philippines (2021). Field data shows 6,800-fold increase in endosperm carotenoids vs wild type. VAD (vitamin A deficiency) affects 190 million children globally.

Herbicide-Resistant EPSPS (Roundup Ready)

5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) catalyses the penultimate step of the shikimate pathway, essential for aromatic amino acid (Phe, Tyr, Trp) synthesis. Glyphosate (Roundup) is a competitive inhibitor of EPSPS (vs PEP substrate). A single substitution Pro101Ser (CP4 EPSPS) from Agrobacterium sp. CP4 reduces glyphosate binding 10,000-fold while retaining catalytic activity. Expressed under the 35S or FMV promoter with a chloroplast-targeting signal peptide.

\( \text{shikimate-3-P} + \text{PEP} \xrightarrow{\text{EPSPS (Pro101Ser mutation)}} \text{EPSP} + \text{P}_i \quad [\text{glyphosate-resistant}] \)

Enhanced Oil Content (FAD2 Engineering)

Silencing FAD2 (delta-12 fatty acid desaturase) in soybean increases oleic acid (C18:1) from ~20% to >80% at the expense of polyunsaturated fatty acids. High-oleic oils have improved oxidative stability for frying. CRISPR-edited high-oleic soybeans (Calyxt, 2019) are the first food product from CRISPR editing to reach market.

CRISPR/Cas9 for Pathway Editing

Single guide RNA (sgRNA) directs Cas9 to the target locus for precise cleavage. Applications: knock-out of catabolism genes (e.g., lycopene β-cyclase to accumulate lycopene); base editing for herbicide resistance; prime editing for precise point mutations; CRISPRa/i for transcriptional modulation without sequence changes.

\( \text{sgRNA} + \text{Cas9} \rightarrow \text{DSB at PAM site} \rightarrow \text{NHEJ (KO) or HDR (knock-in)} \)

Metabolic Flux Balance Analysis (FBA)

FBA models a metabolic network as a stoichiometric matrix S (metabolites × reactions) and finds the flux vector v that satisfies the steady-state assumption while maximising a biological objective (e.g., biomass production):

FBA Mathematical Framework

\( \text{Maximise: } c^T \mathbf{v} \)

\( \text{Subject to: } \mathbf{S} \cdot \mathbf{v} = \mathbf{0} \)

\( v_{\min,i} \le v_i \le v_{\max,i} \)

The steady-state constraint S·v = 0 enforces mass balance for every metabolite. Solved via linear programming (LP). The objective c^T is typically the biomass reaction.

Genome-Scale Models (GEMs)

Plant GEMs include Arabidopsis thaliana AraGEM (~1,400 reactions), wheat, maize, tomato, and cyanobacteria models. They integrate gene-reaction associations (GPR rules) and compartment specificity. Used to:

Identify essential genes (lethal knockouts)
Predict metabolic engineering targets
Simulate phenotypes under nutrient limitation
Integrate omics data (GIMME, IMAT methods)

Python: Flux Balance Analysis (FBA) on a Toy Network

Build a 5-metabolite, 7-reaction stoichiometric network, assemble the matrix S, and solve FBA via linear programming to maximise biomass flux. Then explore the trade-off between byproduct secretion and biomass yield.

Python

script.py141 lines

import numpy as np
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt

# Simple flux balance analysis (FBA) for a toy plant metabolic network
# Network: 5 metabolites (A,B,C,D,E), 7 reactions
#
# Reactions:
# R1: 0 -> A          (uptake/input, glucose entry)
# R2: A -> B + C      (glycolysis split)
# R3: B -> D          (TCA branch)
# R4: C -> D          (amino acid synthesis branch)
# R5: D -> E          (biomass precursor)
# R6: B -> 0          (secretion / fermentation)
# R7: E -> 0          (biomass formation, objective)
#
# Stoichiometric matrix S (metabolites x reactions):
# Rows: A, B, C, D, E
# Cols: R1, R2, R3, R4, R5, R6, R7

S = np.array([
    # R1  R2  R3  R4  R5  R6  R7
    [  1, -1,  0,  0,  0,  0,  0],   # A: produced by R1, consumed by R2
    [  0,  1, -1,  0,  0, -1,  0],   # B: produced by R2, consumed by R3, R6
    [  0,  1,  0, -1,  0,  0,  0],   # C: produced by R2, consumed by R4
    [  0,  0,  1,  1, -1,  0,  0],   # D: produced by R3+R4, consumed by R5
    [  0,  0,  0,  0,  1,  0, -1],   # E: produced by R5, consumed by R7 (objective)
], dtype=float)

print("Stoichiometric matrix S (5 metabolites x 7 reactions):")
print(S)
print()

# FBA: maximise v7 (biomass) subject to S*v = 0 and bounds
# We solve with linear programming using scipy
from scipy.optimize import linprog

n_rxns = S.shape[1]
# Objective: maximise v7 = minimise -v7
c = np.zeros(n_rxns)
c[6] = -1.0  # minimise -v7 (i.e. maximise v7)

# Equality constraint: S * v = 0
A_eq = S
b_eq = np.zeros(S.shape[0])

# Bounds: all fluxes >= 0 (irreversible), R1 <= 10 (substrate input limit)
bounds = [(0, 10) if i == 0 else (0, None) for i in range(n_rxns)]

result = linprog(c, A_eq=A_eq, b_eq=b_eq, bounds=bounds, method='highs')

if result.success:
    v_opt = result.x
    print("FBA solution found!")
    rxn_names = ['R1 (uptake)', 'R2 (glycolysis)', 'R3 (TCA)', 'R4 (amino acid)', 'R5 (precursor)', 'R6 (secretion)', 'R7 (biomass)']
    for name, flux in zip(rxn_names, v_opt):
        print(f"  {name}: {flux:.4f}")
    print(f"  Maximum biomass flux (v7): {v_opt[6]:.4f}")
    print(f"  Carbon yield: {v_opt[6] / v_opt[0] * 100:.1f}%")
else:
    print("FBA infeasible:", result.message)
    v_opt = np.zeros(n_rxns)

# Now explore: effect of R6 (secretion/byproduct) flux on biomass yield
R6_upper_bounds = np.linspace(0, 10, 50)
biomass_yields = []
for R6_max in R6_upper_bounds:
    bnds = [(0, 10), (0, None), (0, None), (0, None), (0, None), (0, R6_max), (0, None)]
    res = linprog(c, A_eq=A_eq, b_eq=b_eq, bounds=bnds, method='highs')
    if res.success:
        biomass_yields.append(res.x[6])
    else:
        biomass_yields.append(0)

fig, axes = plt.subplots(1, 3, figsize=(14, 5))
fig.patch.set_facecolor('#0f172a')
for ax in axes:
    ax.set_facecolor('#1e293b')
    ax.tick_params(colors='#94a3b8')
    ax.spines['bottom'].set_color('#475569')
    ax.spines['left'].set_color('#475569')
    ax.spines['top'].set_visible(False)
    ax.spines['right'].set_visible(False)

# Plot 1: Optimal flux distribution
colors_bars = ['#a3e635', '#34d399', '#22d3ee', '#60a5fa', '#a78bfa', '#f472b6', '#fbbf24']
short_names = ['R1
(uptake)', 'R2
(glycol.)', 'R3
(TCA)', 'R4
(AA)', 'R5
(prec.)', 'R6
(secret.)', 'R7
(biomass)']
bars = axes[0].bar(short_names, v_opt, color=colors_bars, edgecolor='#475569', alpha=0.85)
axes[0].set_ylabel('Flux (mmol/gDW/h)', color='#94a3b8', fontsize=9)
axes[0].set_title('FBA: Optimal Flux Distribution', color='#e2e8f0', fontsize=10)
for bar, val in zip(bars, v_opt):
    axes[0].text(bar.get_x() + bar.get_width()/2, bar.get_height() + 0.05,
                 f'{val:.2f}', ha='center', va='bottom', color='#e2e8f0', fontsize=7)

# Plot 2: Biomass yield vs secretion (byproduct) upper bound
axes[1].plot(R6_upper_bounds, biomass_yields, color='#a3e635', linewidth=2.5)
axes[1].fill_between(R6_upper_bounds, biomass_yields, alpha=0.2, color='#84cc16')
axes[1].set_xlabel('Max allowed secretion flux (R6)', color='#94a3b8', fontsize=9)
axes[1].set_ylabel('Biomass flux (v7)', color='#94a3b8', fontsize=9)
axes[1].set_title('Biomass Yield vs Byproduct Secretion
(FBA trade-off curve)', color='#e2e8f0', fontsize=10)
axes[1].axvline(x=v_opt[5], color='#f59e0b', linestyle=':', alpha=0.7, linewidth=1.5)
axes[1].text(v_opt[5] + 0.2, max(biomass_yields) * 0.5,
             f'Optimal R6={v_opt[5]:.2f}', color='#f59e0b', fontsize=7)

# Plot 3: Metabolic engineering targets bar chart
targets = ['Golden Rice
(PSY+PDS)', 'EPSPS
herbicide res.', 'Enhanced
oils (FAD2)', 'Bt toxin
(cry1Ab)', 'Alkaloid
(STR)', 'Anthocyanin
(CHS/DFR)']
fold_change = [6800, 1.0, 3.5, 1e5, 12, 25]
bar_colors2 = ['#fbbf24', '#f59e0b', '#34d399', '#a3e635', '#ec4899', '#a78bfa']
y_pos = np.arange(len(targets))
axes[2].barh(y_pos, [np.log10(max(f, 1)) for f in fold_change], color=bar_colors2, alpha=0.85, edgecolor='#475569')
axes[2].set_yticks(y_pos)
axes[2].set_yticklabels(targets, fontsize=7.5, color='#94a3b8')
axes[2].set_xlabel('log10(fold improvement)', color='#94a3b8', fontsize=9)
axes[2].set_title('Key Plant Metabolic Engineering Achievements', color='#e2e8f0', fontsize=10)

for i, (f, name) in enumerate(zip(fold_change, targets)):
    axes[2].text(np.log10(max(f, 1)) + 0.02, i, f'x{f}', va='center', color='#e2e8f0', fontsize=7)

fig.suptitle('Metabolic Engineering & Flux Balance Analysis in Plants',
             color='#e2e8f0', fontsize=12, fontweight='bold')
plt.tight_layout()
plt.savefig('output.png', dpi=110, bbox_inches='tight',
            facecolor='#0f172a', edgecolor='none')
print()
print("S * v_opt (should be ~0):", np.round(S @ v_opt, 6))

Click Run to execute the Python code

Code will be executed with Python 3 on the server

← Ch 19: Stress Biochemistry Ch 21: Systems Plant Biology →